Multistep Fractionation and Mass Spectrometry Reveal Zwitterionic and Anionic Modifications of the N- and O-glycans of a Marine Snail*

Various studies in the past have revealed that molluscs can produce a wide range of rather complex N-glycan structures, which vary from those occurring in other invertebrate animals; particularly methylated glycans have been found in gastropods, and there are some reports of anionic glycans in bivalves. Due to the high variability in terms of previously described structures and methodologies, it is a major challenge to establish glycomic workflows that yield the maximum amount of detailed structural information from relatively low quantities of sample. In this study, we apply differential release with peptide:N-glycosidases F and A followed by solid-phase extraction on graphitized carbon and reversed-phase materials to examine the glycome of Volvarina rubella (C. B. Adams, 1845), a margin snail of the clade Neogastropoda. The resulting four pools of N-glycans were fractionated on a fused core RP-HPLC column and subject to MALDI-TOF MS and MS/MS in conjunction with chemical and enzymatic treatments. In addition, selected N-glycan fractions, as well as O-glycans released by β-elimination, were analyzed by porous graphitized carbon-LC-MS and MSn. This comprehensive approach enabled us to determine a number of novel modifications of protein-linked glycans, including N-methyl-2-aminoethylphosphonate on mannose and N-acetylhexosamine residues, core β1,3-linked mannose, zwitterionic moieties on core Galβ1,4Fuc motifs, additional mannose residues on oligomannosidic glycans, and bisubstituted antennal fucose; furthermore, typical invertebrate N-glycans with sulfate and core fucose residues are present in this gastropod.

Various studies in the past have revealed that molluscs can produce a wide range of rather complex N-glycan structures, which vary from those occurring in other invertebrate animals; particularly methylated glycans have been found in gastropods, and there are some reports of anionic glycans in bivalves. Due to the high variability in terms of previously described structures and methodologies, it is a major challenge to establish glycomic workflows that yield the maximum amount of detailed structural information from relatively low quantities of sample. In this study, we apply differential release with peptide:Nglycosidases F and A followed by solid-phase extraction on graphitized carbon and reversed-phase materials to examine the glycome of Volvarina rubella (C. B. Adams, 1845), a margin snail of the clade Neogastropoda. The resulting four pools of N-glycans were fractionated on a fused core RP-HPLC column and subject to MALDI-TOF MS and MS/MS in conjunction with chemical and enzymatic treatments. In addition, selected N-glycan fractions, as well as O-glycans released by ␤-elimination, were analyzed by porous graphitized carbon-LC-MS and MS n . This comprehensive approach enabled us to determine a number of novel modifications of protein-linked glycans, including N-methyl-2-aminoethylphosphonate on mannose and N-acetylhexosamine residues, core ␤1,3-linked mannose, zwitterionic moieties on core Gal␤1,4Fuc motifs, additional mannose residues on oligomannosidic glycans, and bisubstituted antennal fucose; furthermore, typical invertebrate N-glycans with sulfate and core fucose residues are present in this gastropod. Molluscs represent one of the largest groups of animals on the planet; there is an estimated 200,000 species, which vary in morphology from gastropods (snails) through to cephalopods (octopus) and live in a range of marine, aquatic, and terrestrial environments (1). Many molluscs are familiar due to their shells or being seafood. Less appreciated is perhaps their ecological role as filter feeders or scavengers and their being an indicator for water quality (2)(3)(4); also, some molluscs are intermediate hosts for pathogens such as viruses or schistosomes (5,6).
In glycobiological terms, the most studies on molluscs have been structural characterizations of the N-glycans on hemocyanins of a range of gastropods, such as from keyhole limpet (Megathura crenulata; KLH is an often-used carrier protein for immunization), Lymnaea stagnalis, Helix pomatia, and Rapana venosa (7)(8)(9)(10). Furthermore, glycans from cephalopod rhodopsins, proteins of bivalves involved in biomineralization, or whole snail viscera have also been analyzed. Including our recent study on the hemocytes and plasma of the eastern oyster (Crassostrea virginica), the variety of modifications of N-glycans in these organisms is immense and includes branched fucose residues, glucuronylation, sulfation, methylation, core xylose, and galactosylation of core fucose as well as LacdiNAc and blood-group-like motifs (11)(12)(13)(14)(15). On the other hand, there is only scattered information regarding the biosynthesis of mollusc N-glycan epitopes, based on assay of some fucosyl-, xylosyl-, N-acetylglucosaminyltransferases, and N-acetylgalactosaminyltransferases (16 -18); also, probably only two mollusc glycosyltransferases have ever been characterized in recombinant form (19,20).
The high variability and lack of predictability of mollusc glycomes mean that a suitable glycomic workflow has to be employed that takes account of the maxim "expect the unexpected." Thereby, in comparison to mammalian glycomes with known major components, the analyses of those of lower eukaryotes can present major challenges. In the past, mollusc glycans from either a single glycoprotein or from tissue were very often analyzed in any single study by one or two methods (e.g. GC-MS and NMR or MALDI-TOF MS/MS of HPLC-frac-tionated N-glycans, LC-MS/MS of glycopeptides, or GC-MS and MS n of permethylated N-glycans; see references above). In some cases, chemical and enzymatic treatments were employed. Here, we have sought to maximize the potential of off-line MALDI-TOF MS and MS/MS by prefractionating Nglycans first on the basis of whether they can be released by peptide:N-glycosidase A or F (the former being able to remove glycans containing core ␣1,3-fucose (21)) and then using solid-phase extraction on nonporous graphitized carbon (for an initial separation of anionic from neutral glycans (22)) and on a reversed-phase resin (which aids enrichment of glycans with substitutions of core ␣1,6-fucose). Subsequent use of a fused core reversed-phase (RP)-HPLC column (23) resulted in high-resolution separation into fractions containing either a single or very few glycan species that facilitated further MS-based analyses; as this RP column offers isomeric/isobaric separation, HPLC fractionation was a prerequisite for the definition of the individual N-glycan structures. Furthermore, the residual glycopeptides (posttreatment with peptide:N-glycosidases) were subject to ␤-elimination to release the O-glycans followed by LC-MS.
On the basis of these considerations, we have examined the N-and O-glycomes of a margin snail (Volvarina rubella), a species of carnivorous and scavenging marine gastropod first described as Marginella rubella in 1845 (24). Using off-line LC-MALDI-TOF MS and on-line LC-ESI-MS, we reveal a particularly complex N-glycome encompassing a range of oligomannosidic, paucimannosidic, core-modified, and complex (up to triantennary) N-linked oligosaccharides with also a number of anionic and zwitterionic modifications, which are also present on Oglycans. Although some of these features are also found on N-glycans or lipid-linked glycans of other species, the majority of the ϳ100 structures are described here for the first time.

EXPERIMENTAL PROCEDURES
Biological Material and Glycan Preparation-Adult margin snails, cultivated on a shrimp-based diet, were obtained from the "Haus des Meeres-Aqua Terra Zoo," a public aquarium in Vienna, Austria. The species was determined as being V. rubella (C. B. Adams, 1845 Mollusca: Gastropoda: Muricoidea: Marginellidae) (25,26). The snails (3.4 g wet weight; 1.2 g after lyophilization) were washed once with deionized water and stored at Ϫ80°C. After thawing, the material was heat inactivated for 10 min in boiling water and lyophilized prior to grinding in liquid nitrogen. The powder was suspended in deionized water and the pH adjusted with ammonium carbonate buffer to 8.2. CaCl 2 was added to a final concentration of 0.5 mM before addition of 2 mg Thermolysin (Promega, Madison, WI). Proteolysis was allowed to proceed for 2 h at 70°C.
The glycopeptides were purified using standard laboratory protocols (27) prior to release with PNGase F (Roche) overnight at 37°C. After cation exchange chromatography on Dowex, the unbound material was subject to solid-phase extraction on nonporous graphitized carbon (NPGC; ENVIcarb, Supelco, Bellefonte, PA) and eluted with 40% acetonitrile 1 or 40% acetonitrile with 0.1% TFA. A further solid-phase extraction step was performed on a C18 reversed-phase resin (LiChroprep; Merck, Darmstadt, Germany), and the glycans were eluted with water and with stepwise increases in the methanol concentration (15%, 40%, 100% (v/v)). Glycan-containing fractions, as judged by MALDI-TOF MS, were fluorescently labeled with 2-aminopyridine. The remaining glycopeptides were gel filtrated (Sephadex G25) prior to incubation with PNGase A (recombinant form prepared in house) overnight at 37°C. The glycans released with this enzyme underwent the same purification and fluorescent-labeling steps as for the PNGase F released ones; however, no anionic glycans were detected in this pool. The residual glycopeptides after PNGase A digestion were subjected to reductive ␤-elimination with 0.5 M NaBH 4 and 50 mM NaOH at 50°C overnight (see Fig. 1A for workflow summary). Released O-glycans were cleaned up as previously described prior to LC-MS/MS analysis (28).
N-glycan Fractionation-The pyridylaminated N-glycans were fractionated by HPLC using a Kinetex 5 m RP-column (XB-C18 100A, 250 ϫ 4.6 mm; Phenomenex®, Torrance, CA), and a gradient of methanol in 0.1 M ammonium acetate, pH 4, up to 16.5% over 44 min was applied at a flow rate of 0.8 ml/min as follows: 0 -30 min, 0 -9% methanol; 30 -35 min, 9 -12% methanol; 35-40 min, 12-16.5% methanol; 40 -44 min, 16.5% methanol; and 44 -50 min, return to 0% methanol. Lyophilized HPLC fractions were dissolved in water and subject to MALDI-TOF MS. A partial dextran hydrolysate (Sigma-Aldrich, St. Louis, MO; subsequently pyridylaminated) was used to calibrate the RP-HPLC column in terms of glucose units (g.u.) as previously described (29); the masses of selected peaks of this linear glucose polymer were verified by MALDI-TOF MS after RP-HPLC (e.g. 6 and 10 g.u. eluting at 20 and 26 mins; m/z 1069.6 and 1717.8 as MALDI-TOF Mass Spectrometry-Free glycans and pyridylaminated glycans were analyzed in positive and negative-ion modes using either Bruker Autoflex Speed or UltrafleXtreme instruments (both with 1000 Hz Smartbeam™-II lasers) with 6-aza-2-thiothymine as matrix; calibration was performed using a Bruker peptide standard. MS/MS was performed by laser-induced dissociation (precursor ion selector was generally Ϯ0.6%). The detector voltage was normally set at 1977 V for MS and 2133 V for MS/MS; 1000 -2000 shots from different regions of the sample spots were summed. Spectra were processed with the manufacturer's software (Bruker Flexanalysis 3.3.80) using the SNAP algorithm with a signal/noise threshold of 6 for MS (unsmoothed) and 3 for MS/MS (four-times smoothed). Glycan MS and MS/MS spectra (ϳ2700 in total) were manually interpreted on the basis of the masses of the predicted component monosaccharides, the differences of mass in glycan series, fragmentation patterns, and results of enzymatic and chemical treatments. For the ϳ100 structures (see the Supplemental Table 1 for a summary of evidence for each glycan), the minimum criterion for inclusion in the Supplemental Table 1 was an interpretable MALDI-TOF MS/MS spectrum. Furthermore, examples for each core and antennal motif were verified by digestion data; comparison was also made to elution, in terms of glucose units, with previous data on glycans of other species. About 20 representative structures are also supported by LC-MS n data. Calculated theoretical masses were verified using Glyco-Workbench 2.0. The deviation between calculated and observed m/z values was typically 0.1-0.2 Da.
LC-ESI Mass Spectrometry-Selected RP-HPLC fractions of pyridylaminated N-glycans were also analyzed by online LC-MS/MS using a 10 cm ϫ 150 m inner diameter column, prepared in-house, containing 5 m porous graphitized carbon (PGC) particles coupled to an LTQ ion trap mass spectrometer (Thermo Scientific, Waltham, MA). Glycans were eluted using a linear gradient from 0 -40% acetonitrile in 10 mM ammonium bicarbonate over 40 min at a flow rate of 10 l/min. The eluted N-glycans were detected in negative-ion mode with an electrospray voltage of 3.5 kV, capillary voltage of -33.0 V, and capillary temperature of 300°C. Specified ions were isolated for MS n fragmentation by collision-induced dissociation with the collision energy set to 30%. Air was used as a sheath gas, and mass ranges were defined dependent on the specific structure to be analyzed. The data were processed using the Xcalibur software (version 2.0.7, Thermo Scientific). Glycans were identified from their MS/MS spectra by manual annotation according to the nomenclature of Domon and Costello (30).
For removal of methylaminoethylphosphonate, phosphorylcholine, or branched fucose residues, selected fractions were dried and incubated for 48 h at 0°C with 3 l 48% (v/v) hydrofluoric acid prior to evaporation in a centrifugal concentrator (33). The samples were diluted in water and re-evaporated, before redissolving once again. The chemically or enzymatically treated fractions were subject to MALDI-TOF MS and MS/MS (as above) without further purification.

RESULTS
Overall N-glycome of a Marine Margin Snail-The analytical strategy for V. rubella was based on selective release and fractionation of different classes of N-glycans from thermolysin-generated glycopeptides. We, thereby, modified a procedure we previously used for Dictyostelium N-glycans (22). Using two different peptide:N-glycosidases (PNGase F then PNGase A; the latter able to remove glycans with core ␣1,3fucose) and by solid-phase extraction on nonporous graphitized carbon and reversed-phase resin, we obtained a total of four different pools of N-glycans (PNGase A-released and "neutral," "hydrophobic," and "anionic-enriched" PNGase Freleased fractions; see also Fig. 1A). The free glycan pools were separately fluorescently labeled with 2-aminopyridine prior to MALDI-TOF MS and HPLC.
The first impression of the PNGase A and neutral-enriched PNGase F digests (both eluted with acetonitrile from the graphitized carbon and with water from the C18 column) is of a set of paucimannosidic and oligomannosidic N-glycans typical for invertebrates (Hex 2-9 HexNAc 2 and Hex 1-3HexNAc 2 Fuc 1-2 ; Figs. 1B and 1D). The PNGase A pool contains relatively low amounts of difucosylated species (m/z 1119 and 1281 as [MϩH] ϩ ); preliminary MS/MS of glycans in the entire pool, confirmed later on individual fractions, showed the occurrence of m/z 592 Y 1 -fragments suggestive for disubstitution of the core GlcNAc by both ␣1,3 and ␣1,6-fucose. Additionally, the overall mass spectra of both of these pools indicate the presence of glycans whose masses (m/z 948, 1272, and 1555 as [MϩH] ϩ ) could not be accounted for by typical monosaccharide compositions. Only in later experiments did it become clear that these are carrying the zwitterionic N-methyl-2-aminoethylphosphonate moiety (⌬m/z 121).
The hydrophobic neutral PNGase F-released glycans (eluted with 15% methanol from the C18 solid-phase material) were primarily fucosylated (Fig. 1C). Unexpected also, difucosylated glycans were present in this pool (also of m/z 1119 and 1281 as [MϩH] ϩ ); preliminary MS/MS, however, showed the presence of m/z 754 Y 1 -fragments indicative of a structural difference as compared with the isomeric glycans in the PNGase A pool. Two unusual difucosylated glycans with zwitterionic modifications (as later determined; m/z 1402 and 1446 as [MϩH] ϩ ) were also obvious in this pool.
Finally, the anionic-enriched PNGase F-released pool that was eluted with acetonitrile supplemented with 0.1% trifluoroacetic acid from graphitized carbon and then with water from the C18 column contained two basic types of N-glycans ( which, as these are anionic species easily detected in the negative mode, could correspond to a composition of Hex 1 Me 1 Fuc 1 HexA 1 . In order to reveal more about the structures of the glycans in this and the other pools, they were applied to a fused core reversed-phase HPLC column (Kinetex XB-C18), which was chosen due to its increased resolution as compared with other RP-HPLC columns in a recent study (33). All fractions were then subject to MALDI-TOF MS and MS/MS analyses in conjunction with chemical and enzymatic treatments; example structures were also analyzed by LC-MS n . Some of the observed structures can, in terms of elution time (glucose units, g.u.), fragmentation patterns, and digestion data (see also Supplemental Table), be directly compared with those in our recent study on N-glycans of the nematode Pristionchus pacificus (33). Thereby, the ability of RP columns to distinguish isomers of oligomannosidic or core fucosylated glycans or to separate glycans that only differ in terms of which antenna is elongated is in accordance to previously published data (22,29,34,35). In the case of acidic glycans, the decrease in RP-HPLC retention time correlates with the number of anionic moieties (36,37).
Smaller Mono-and Difucosylated N-glycans-As mentioned above, a number of lower molecular mass glycans carrying one or two fucose residues were observed in the overall MALDI-TOF MS spectra of the PNGase A (neutral) and PNGase F (hydrophobic) pools. RP-HPLC of the PNGase A pool resulted in separation of glycans across a range of elution times ( Fig. 2A); the early eluting monofucosylated species (5.2-5.5 g.u.; Hex 2-3 HexNAc 2 Fuc 1 ; m/z 973 and 1135 as [MϩH] ϩ ) were concluded to be core ␣1,3-fucosylated due to the comparison to Pristionchus N-glycans (33) and the general trend for this modification to result in lower retention on RP-HPLC columns (35). The most dominant peak on the RP-HPLC of this pool (8.4 g.u.) contained difucosylated glycans (Hex 2-3 HexNAc 2 Fuc 2 ; m/z 1119 and 1281), which yielded MS/MS spectra showing the fragment of m/z 592, Fuc 2 GlcNAc 1 -PA, being indicative of difucosylation of the core GlcNAc-PA (Fig. 3A). Furthermore, the elution time was compatible to previous data for such glycans on the same column (33) with a retention time between those of glycans with solely core ␣1,3or ␣1,6-fucose (38); the linkage of the mannose residues was verified using specific ␣1,2/3or ␣1,6mannosidases (see Supplemental Table). As expected from the solid-phase C18 elution properties, the hydrophobic PNGase F-released pool contained rather late-eluting glycans (9 g.u. and above; Fig. 2B). A minor amount of monofucosylated glycans carrying core ␣1,6-fucose was present; among these, fragmentation resulted not only in observation of m/z 446 (Fuc 1 GlcNAc 1 -PA) but also of m/z 608 fragment ions (Hex 1 Fuc 1 GlcNAc 1 -PA; Fig. 3B) of a type observed with nematode N-glycans previously shown to have ␤1,4-galactosylated core ␣1,6-fucose (33, 39) and verified by glycosidase digestions (see Supplemental Table). The tendency of core ␣1,6-fucose to result in late elution on RP-HPLC first noted by Tomiya et al. (29) and also observed for in vitro products of the relevant core fucosyltransferase (40) is even more pronounced for glycans with galactose substitutions of the fucose residue.
The majority of glycans, though, in the hydrophobic pool were difucosylated. However, as this is a PNGase F-released pool, the typical core difucosylation was not expected. Indeed, MS/MS of these glycans did not reveal the presence of any m/z 592 fragments (Fuc 2 GlcNAc 1 -PA; see above): on the other hand, fragments of m/z 754 (Figs. 3C, 3D, and 3F), reminiscent of ones previously found in Caenorhabditis elegans (39), suggested that core ␣1,6-fucose was capped with a hexose and a fucose. In the case of Hex 3 HexNAc 2 Fuc 2 -PA (Ͼ20 g.u.; m/z 1281), use of a specific mannosidase proved that the single ␣-mannose is 1,6-linked and led to loss of the m/z 827 MS/MS fragment ( form of Hex 3 HexNAc 2 Fuc 2 is concluded to carry a Fuc␣1,2Gal␤1,4Fuc␣1,6-modification on the core proximal GlcNAc residue. Some glycans in the anionic pool with antennal modifications also possessed these extended core fucose modifications (see below).
Five N-glycans in the hydrophobic pool had compositions that suggested the presence of a nonsugar modification. The mass difference of 165 Da between the glycans of m/z 1284 and 1446 as compared with those of m/z 1119 and 1281 was suggestive for the presence of phosphorylcholine, which is also a known modification of nematode N-glycans (33). Our standard procedure to analyze glycans modified with a phosphodiester is to treat them with hydrofluoric acid (22); in the case of the m/z 1446 glycan (Hex 3 HexNAc 2 Fuc 2 PC 1 -PA; 13.7 g.u.), there was indeed a loss of 165 Da, and the unusual fragments of m/z 474 and 620 (interpreted as being Standard Paucimannosidic Glycans-The largest pool of glycans in terms of total fluorescence was of the neutralenriched PNGase F digest. RP-HPLC of this pool resulted in some 40 glycan-containing fractions (Fig. 6). The later-eluting RP-HPLC fractions (Ͼ11 g.u.) contain some typical monofucosylated species (Hex 1-3 HexNAc 2-4 Fuc 1 -PA), which display major MS/MS protonated fragments at m/z 446 (GlcNAc 1 Fuc 1 -PA). The elution times are indicative of ␣1,6-fucosylation as compared with previous studies using the same column and gradient (33) and literature values for such glycans (29). Based on its retention time, the major m/z 973 glycan (16 g.u.) was assumed to be Man␣1,6Man␤1,4GlcNAc␤1,4(Fuc␣1,6)GlcNAc-PA.
Supporting this assignment, ␣1,6-mannosidase and bovine ␣-fucosidase removed one residue each to result in products of m/z 811 and 827, respectively (see Supplemental  Table). Furthermore, there are minor amounts of Hex 3 HexNAc 3 Fuc 0 -1 -PA glycans (m/z 1192 and 1338) with the third HexNAc residue on either the ␣1,3or ␣1,6-mannose residue as judged by the retention times in comparison to previous studies (33) as well as the ␣1,2/3-mannosidase  Supplemental  Table). The contrasting retention times resulting from extension of the different antennae are exemplified (red arrows, Fig. 6A) for the glycans of m/z 1192, 1338, and 1395, with elongation of the ␣1,3-arm causing earlier elution as compared with the effect of substitution of the ␣1,6-arm (41).
Typical also of invertebrates are nonfucosylated paucimannosidic glycans such as Man 2-5 GlcNAc 2 ; other than m/z 827 glycan (Man 2 GlcNAc 2 -PA; at 9.8 g.u.), being also sensitive to ␣1,6-mannosidase and so lacking the core ␣1,3-mannose, additional glycans of this class were of rather low amounts as compared with insects and nematodes. On the other hand, one of the three major peaks contained a glycan with m/z 1313 (Hex 5 HexNAc 2 -PA) at an unusual elution position (7.2 rather than 8.5 g.u. for the standard isomer) and was proven by LC-MS n to also have no lower arm ␣1,3-mannose (see below). The tendency for glycans of this gastropod to lack this residue is also shared with monoantennary glycans with "complex" antennae (e.g. m/z 1528 and 1674 in the anionic pool; see below).
Unusual Isomers of Oligomannosidic Glycans-In addition to fractions in the neutral-enriched PNGase F pool (4 -9 g.u.; Fig. 6A) containing the normal oligo-and paucimannosidic glycans (Hex 3-9 HexNAc 2 -PA; m/z 989 -1961, [MϩH] ϩ ), as based on their coelution with known oligomannosidic glycans and similar fragmentation (see Supplemental Table), the multiplicity of further elution positions for these masses raised suspicions as to the exact compositions and structures of many of these glycans. For instance, there were three elution positions for glycans with m/z 1799 (Hex 8 HexNAc 2 -PA; 4.7, 5.0, and 5.9 g.u.). The 5.0 and 5.9 g.u. elution positions would correspond to the standard Man8B and Man8A isomers (respectively, with the "middle" or "lower" mannoses of Man9 having been removed by endoplasmic reticulum mannosidase or by a Golgi mannosidase I) as found in Dictyostelium or Pristionchus (22,33). The earlier eluting form of Hex 8 HexNAc 2 -PA (4.7 g.u.), however, does not correspond to any standard isomer. After proving loss of seven residues with jack bean ␣-mannosidase, this glycan was digested with linkage-specific mannosidases. Surprisingly, the Xanthomonas ␣1,6-mannosidase removed one mannose, re-  Fig. 7A). Based on the specificity of this enzyme (42) and our experience that it otherwise does not digest any standard oligomannosidic glycan without pretreatment with other specific mannosidases (43), it is concluded that the 4.7 g.u. glycan has a terminal ␣1,6-mannose linked to an unsubstituted mannose. GlcNAc 2 isomers before and after digestion with specific mannosidases; the coeluting Man 6 GlcNAc 2 glycan is an earlier-eluting epimer (core ManNAc-PA as reduction artifact) of the structure in the major 7.2 g.u. fraction.
For an aberrantly eluting form of Hex 7 HexNAc 2 -PA (m/z 1637; 5.8 g.u.), treatment with both ␣1,6or ␣1,2-specific mannosidases was performed. While both mannosidases removed one mannose from the glycan (Figs. 7B and 7C), the shift in the major MS/MS Y3␤-fragment from m/z 989 to m/z 827 was only observed with the ␣1,2-specific mannosidase. These data were considered to indicate that the accessible ␣1,6-mannose linked to an unsubstituted mannose was on the upper or middle arms, whereas the ␣1,2-mannose is on the lower arm; this would rule out that the ␣1,6-mannose is a lower arm "outer chain" modification as in fungi (43). Final examples of the unusual oligomannosidic glycans are forms of Hex 5-6 HexNAc 2 (7.2 g.u.), which were also susceptible to ␣1,6-mannosidase digestion and shifted to coeluting with known Man 4 -5 GlcNAc 2 -PA (two peaks of around 9 g.u. with no alteration in the major MS/MS fragment; Fig. 7D). The Hex 5-6 HexNAc 2 glycans were, on the other hand, resistant to ␣1,2-mannosidase but lost in total four or five residues upon jack bean ␣-mannosidase digestion; however, only the Hex 6 HexNAc 2 was sensitive to ␣1,2/3-mannosidase, compatible with its fragmentation showing the presence of a lower arm (see Supplemental Table).
In order to locate the extra ␣1,6-mannose residue more exactly, LC-ESI-MS/MS in the negative-ion mode was performed on one of these glycans (Man 5 GlcNAc 2 -PA; [M-H] Ϫ ions at m/z 1311; Fig. 7E), which is proposed to have an ␣1,6-mannose linked to the middle or upper arm and be devoid of the lower arm (␣1, 3-mannose). Fragmentation ions at m/z 869 and 1012 were assigned, respectively, as products of the 2,4 A cross-ring cleavage of the penultimate GlcNAc ( 2,4 A 5 ) and glycosidic cleavage within the chitobiose (B 5 ). Cross-ring cleavages of ␤-mannose ( 0,3 A 4 ions at m/z 737 and 0,4 A 4 ions at m/z 707) together with "D ions" at m/z 809/791 (mass of 6-antenna plus ␤-Man (44) indicate that this N-glycan is devoid of a lower arm but has four mannoses on the upper arm. The presence of B 3 /C 3 ions at m/z 665/647 but the lack of ions at m/z 503 (mass of three mannoses plus H 2 O) suggests that the four mannoses are branched rather than linear. The presence of fragment ions at m/z 485 but none at m/z 467 suggests that two mannoses are linked to the C3 of the ␣1,6-mannose (also confirmed by MS 3 , data not shown) while only one mannose is linked to the C6 of the ␣1,6mannose. Thus, this verifies that the free ␣1,6-mannose is on the middle (B) arm. Analogously, a form of Man 4 GlcNAc 2 as well as another Man 5 GlcNAc 2 isomer also possessed this "middle" ␣1,6-mannose as judged by ␣1,6-mannosidase digests (see Supplemental Table).
A further unusual oligomannosidic modification is the presence of an additional ␣1,3-mannose on the lower arm of two early-eluting Man 8 -9 GlcNAc 2 isomers. Specific mannosidases were employed to serially digest these structures. First Aspergillus ␣1,2-mannosidase removed one or two residues with a shift in the fragmentation pattern and ␣1,6-mannosidase a further residue (Figs. 7F-7H). Finally, ␣1,2/3-mannosidase treatment resulted in a product of m/z 1151 (Fig. 7I) displaying a major Y3␤-fragment ion of m/z 665, suggesting that loss of the lower arm only occurred when a mannosidase capable of hydrolyzing ␣1,3-linkages was employed. Mannosylation of the Reducing Terminus-Another unexpected feature of the neutral-enriched PNGase F pool were glycans with major MS/MS Y1-fragment ions at m/z 462 (Hex 1 GlcNAc 1 -PA, Fig. 8); as these glycans contain two core GlcNAc residues, it can be ruled out that this fragment results from an endoglycosidase product. We have previously observed a similar feature for glycans from the eastern oyster and assumed this may be indicative of a hexose attached to the reducing terminal GlcNAc residue, but the low amounts precluded a fuller analysis (12); some of these oyster glycans were also core ␣1,6-fucosylated, so the assumption was that the "extra" core hexose was 1,3-linked. However, due to the larger amounts of such core hexosylated glycans in the margin snail, we were able to test their sensitivity to a number of hexosidases (␣and ␤-galactosidases and ␣and ␤-mannosidases); incubations of an Hex 3 HexNAc 2 -PA (m/z 989 glycan; 11 g.u.) with Xanthomonas ␣1,6-mannosidase and H. pomatia ␤-mannosidase both removed one hexose residue, but only the latter treatment resulted in a loss of the m/z 462 fragment (Figs. 9 A-9C). Thus, we assumed (as the 2 and 4 positions of the proximal core GlcNAc are substituted with an N-acetyl moiety and the distal GlcNAc) that, in analogy to the oyster glycans, this ␤-mannose is 3-linked. In contrast to the oyster, though, the core hexose residue in the snail is present on some hybrid and biantennary glycans lacking core fucose. Interestingly, the effect of core ␤1,3-mannosylation on retention time is relatively minor (a shift of ca. 1 g.u. later than the "parent" structure) as compared with core ␣1,3-fucosylation (ca. 4 g.u. earlier). Furthermore, the presence of the core ␤1,3-mannosylated glycans in the PNGase F pool, as opposed to the absence of core ␣1,3-fucosylated ones, means that the action of PNGase F is not inhibited per se by a 3-substitution of the reducing terminal GlcNAc.
A specific proof for ␤1,3-mannosylation in V. rubella was delivered by LC-ESI-MS/MS data of Hex 3 HexNAc 2 -PA (Fig.  9D). In negative-ion mode, the MS/MS spectra were dominated with Y and Z ions. The presence of Y 1␣ /Z 1␣ ions at m/z 460 and 442, together with B 3␣ ions at m/z 526, indicates that one hexose is linked to the proximal GlcNAc. Fragmentation ions at m/z 600 were interpreted as being derived from 3,5 A cleavage of the proximal GlcNAc, indicating that the hexose is linked to the C3 position of this residue, whereas those at m/z 221 in MS 3 of Y 1␣ are compatible with 2,4 A cleavage of the proximal GlcNAc, which further confirms the C3 substitution of this residue. Taken together, these data indicate that this N-glycan from V. rubella is substituted with ␤1,3-mannose on the proximal core GlcNAc. This modification is also present on some anionic glycans with an antennal branched fucose residue (e.g. m/z 1690 and 1852; see below).
Zwitterionic N-glycans-The neutral-enriched PNGase Freleased pool also contained a number of glycans (see Fig. 6B for the annotated chromatogram) whose compositions and fragmentation patterns did not match known glycans, despite even sharing the same mass with oligomannosidic glycans (e.g. m/z 1151 and 1313 at 14.5 g.u., 1637 at 9.8 g.u., and 1961 at 7.9 g.u.). Even though some individual fragment ions were identical, relatively strong m/z 325 fragments, losses of 121 Da and strong negative-ion mode signals were observed indicative of an unusual modification (Fig. 10). As a modification of 121 Da is unlikely to be a monosaccharide, we considered the possibility that this may be a zwitterionic modification other than phosphoethanolamine (⌬m/z 123) as found in Trichomonas N-glycans (27) or phosphorylcholine (⌬m/z 165) as found in some core-modified glycans described above (see Fig. 5). Based on the literature on invertebrate N-glycans and glycolipids, we considered that this modification is N-methyl-2-aminoethylphosphonate (⌬m/z 121), which has been previously found on glycolipids of one mollusc (45) as well as in unmethylated form (i.e. 2-aminoethylphosphonate, ⌬m/z 107) on locust N-glycans (46). As with standard phosphodiesters, hydrofluoric acid also removes this phosphonate from glycans, resulting in loss of 121 Da and of the m/z 284 and/or 325 fragments (Hex [MAEP] or HexNAc-[MAEP]; see example in Fig. 11B). The treatment appears to leave behind putative monoantennary, pseudohybrid, and biantennary structures. In the case of an m/z 1151 glycan (Hex 2 HexNAc 3 [MAEP]-PA; 14.5 g.u.), incubation in series with hydrofluoric acid and a recombinant branch-specific ␤-Nacetylhexosaminidase did not result in removal of the nonreducing terminal GlcNAc despite hydrolysis of the phosphonate. Due to previous data on the specificity of the insect FDL hexosaminidase (32), this is indicative that the glycan carries the phosphonate-modified GlcNAc on the ␣1,6-arm (Fig.  11C). For the pseudohybrid glycans Hex 3 HexNAc 3 [MAEP] 1-3-PA, the upper arm modification with GlcNAc was indicated by the ability to remove one mannose from the lower arm with the ␣1,2/3-mannosidase after hydrofluoric acid treatment (see Supplemental Table).
In one case of these zwitterionic glycans (m/z 934), we had a modification with aminoethylphosphonate (⌬m/z 107; see  Fig. 10J). In general, the modification with methylaminoethylphosphonate resulted in an earlier retention time than the corresponding neutral "backbone": e.g. compare the biantennary m/z 1395 glycan eluting at 11 g.u. with the forms with two, three, and four methylaminoethylphosphonate residues (m/z 1637, 1758, and 1879) at 9.8, 7.9, and 3.4 g.u. respectively (Fig. 6B and Supplemental Table); the trend to earlier elution on the Kinetex column with an increasing number of zwitterionic residues is shared with phosphorylcholine-modified nematode glycans (33). Further evidence for the modification with methylaminoethylphosphonate and its exact location on mannose and N-acetylglucosamine residues came from LC-ESI-MS/MS data.
In Cross-ring cleavage of ␤-mannose ( 0,2 A 3 ions at m/z 605, 0,4 A 3 ions at m/z 545) suggests that this N-glycan lacks the lower arm, which correlates with the FDL-hexosaminidase insensitivity after hydrofluoric acid treatment. However, MS 3 of B 1 ions (Fig. 12D) supports the presence of substituted GlcNAc rather than substituted mannose and the fragment ions at m/z 221, assigned as 1,3 A cleavage of GlcNAc, indicates that the MAEP is linked to the C3 of GlcNAc. As mentioned above, the 121 Da modification (concluded to be methylaminoethylphosphonate) was also found on galactose of the Fuc␣1,2Gal␤1,4Fuc␣1,6-modification of the core GlcNAc residues (Figs. 3E and 5) while a second isomer of m/z 1402 was modified on the mannose as deduced from the MS/MS (Fig. 3F).
Sulfated N-glycans-The anionic-enriched PNGase F-released glycan pool, accounting for some 15% of the total, contained a range of glycans detected in negative-ion mode, which eluted between 4 and Ͼ20 g.u. (Fig. 13; see also ; the presence of this moiety on the upper arm is supported by specific glycosidase treatments after its removal by hydrofluoric acid (see Supplemental Table).
3-specific ␣-mannosidases and with jack bean hexosaminidase; as only digestion with jack bean mannosidase occurred (Figs. 14E-14H), a modification with sulfated GlcNAc is assumed for the ␣1,3-arm (see also Supplemental Table). Another isomer of Hex 3 HexNAc 3 Fuc 1 S 1 eluted later (15 g.u.) and so is concluded to have the "opposite" configuration (i.e. GlcNAc on the ␣1,6-arm), which would be in keeping with the long-established elution effects of isomeric glycans on RP-HPLC (29). Nonfucosylated and double-sulfated variants were also detected that eluted earlier than the fucosylated monosulfated form (e.g. 6.8 and 6.  (47), sulfation of mannose is known for a lobster N-glycan (48).
Branched Fucose Modifications-In addition to the sulfated glycans, the anionic-enriched pool was rather dominated by glycans with unusual compositions (see discussion of Fig. 1E above), which eluted in a broad range from the RP-HPLC column (Fig. 13). Upon fragmentation in the positive-ion mode, serial loss of 176 and 322 (i.e. 146 ϩ 176) or a loss of 498 (i.e. 322 ϩ 176) en bloc was observed (Fig. 15). In the lower mass range, fragments indicative of the various other aforementioned modifications (core fucose, core mannose, galactosylated core fucose, and methylaminoethylphosphonate; e.g. at m/z 446, 462, 608, and 325, respectively) were present, which suggested that the 498 Da modification (hypothesized to be Hex 1 Me 1 Fuc 1 HexA 1 ) was present on a wide range of glycan types in the margin snail. The efficient ionization in negative mode was considered to be compatible with the presence of a hexuronic acid (176 Da). Interestingly, in the hydrophobic pool, some very late-eluting potential biosynthetic intermediates of these glycans were observed ( Fig.  2B; m/z 1484, 1498, and 1660). Due to no ionization in the negative mode, the high retention on reversed-phase and analyses after hydrofluoric acid (loss of 146 or 322) or bovine ␣-fucosidase (loss of one or two fucose residues; see Sup- plemental Table), these glycans are assumed to lack the glucuronic acid moiety but have either terminal ␣-fucose or methylhexose-substituted ␣-fucose. Therefore, it is also concluded that the second 176 Da unit of the acidic 498 Da modification is a methylated hexose.
As fucose was apparently a component of this modification, we tested the effect of hydrofluoric acid, and indeed, a loss of 498 Da en bloc could be observed (see example treatment of the 12 g.u. fraction; Fig. 16B). The degree of removal (ϳ40%) would be compatible with an ␣1,4-linkage of fucose to the underlying GlcNAc residue; the C2 is occupied by the Nacetyl function, while previous experience with treatment of nematode glycans (49) with hydrofluoric acid indicates that a C3-linked fucose would probably be completely removed and C6-linked fucose not at all. The chemically treated product was then susceptible to the action of the FDL hexosaminidase (Fig. 16C), which is specific, under the conditions used, for removal of GlcNAc from the ␣1,3-arm (32).
Further proof for the proposed position of the branched fucose came from LC-ESI-MS/MS data on two other glycans; the cross-ring cleavage of GlcNAc ( 0,2 A 3 -H 2 O, and 0,2 A 3 ions at m/z 599/617) is compatible with either a 4-or 6-substituted GlcNAc (Fig. 17A). As the bond between Fuc and GlcNAc was sensitive to hydrofluoric acid, we rather conclude that the Fuc residue is linked to C4 of GlcNAc. MS 3 of B 3 ions at m/z 700 yielded ions at m/z 249 ( 0,3 A of Fuc) and 263 ( 2,5 A of Fuc) compatible with one 176 Da unit being linked to C4 of Fuc and another to C2 of Fuc (Fig. 17B). Thus, considering all the evidence, as well as comparisons to the literature on branched fucose in other molluscs (15), a possible model is that the 498 Da modification consists of an ␣1,4-linked fucose modified in turn by a 2-linked methylhexose and a 4-linked glucuronic acid. We further conclude that up to three such units can be present on triantennary glycans (e.g. m/z 3239; Fig. 15L). The definition of the triantennary backbone is based on analogy to the multiantennary glycans with branched fucose found in the blue mussel (15). Some monosubstituted forms were present in the standard "neutral" pool Mucin and O-fucose-Type O-glycans-LC-ESI-MS analysis of the ␤-eliminated O-glycans indicated the presence of a variety of fucose-, sulfate-and (methyl)aminoethylphosphonate-modified di-and trisaccharides (Fig. 18). The majority of the detected structures are, as judged by retention time and fragmentation, based on the mucin core 1 (Gal␤1,3GalNAc) and core 3 (GlcNAc␤1,3GalNAc) structures (50); however, also two O-fucose glycans, of the type familiar from Notch and blood clotting factors (51), were also found. The basic form of core 1 is the most common O-glycan detected in V. rubella; it was also observed in sulfated form (Fig. 18I). Less common were the core 3 structures, but these encompassed fucosylated (HexNAc 2-4 Fuc 1 ), extended (HexNAc 2-3 ), and zwitterionic (HexNAc 3 Me 0 -1 AEP 1-2 ) forms (Figs. 18B-18G). One of the O-fucose glycans was also modified with a zwitterion (AEP-HexNAc-Fuc), which is probably based on GlcNAc␤1,3Fuc (Fig. 18H).  Fig. 13) was treated with hydrofluoric acid (B) and then further incubated with the branch-specific honeybee FDL ␤-hexosaminidase (C) prior to positive-ion mode MALDI-TOF MS. Alterations in the structure were also monitored by MS/MS (see insets for the nondigested and digested forms). About 40% of the antennal modification was removed by hydrofluoric acid (a percentage consistent with ␣1,4-fucosylation) to yield a nonreducing terminal GlcNAc, which is concluded to be on the ␣1,3-arm due to its sensitivity toward the FDL ␤-hexosaminidase.  A, B, and C). Fucosylated HexNAc-rich O-glycans were also found (D and E). As in the case of some Nglycans, O-glycans from the margin snail were also modified with aminoethylphosphonate (AEP) and methylaminoethylphosphonate (MAEP; F and G). In addition, the AEP-HexNAc motif was also found on O-Fuc (panel H). Crossring cleavage of HexNAc ( 0,2 A HexNAc / 0,2 A HexNAc -H 2 O, m/z 226/208) suggested that AEP is linked to either C4 or C6 of HexNAc (H). However, core 1 (not shown) and sulfated core 1 (panel I) were the two most dominant O-glycans isolated from the margin snail.

DISCUSSION
Neutral N-glycans of V. rubella-The data presented here on the margin snail, V. rubella (C. B. Adams, 1845), indicate an unprecedented complexity and variety in an invertebrate Nglycome comprising some 100 structures (see Supplemental  Table). Thereby, some "typical" invertebrate features, such as nonfucosylated paucimannosidic forms, core ␣1,3-fucosylation and core ␣1,3/6-difucosylation are of relatively low abundance and xylosylation or methylation of oligomannosidic glycans, found in terrestial and freshwater gastropods (13), are not at all present. Particularly unusual among the neutral N-glycan modifications of V. rubella is the ␤1,3-mannosylation of the core GlcNAc and the peripheral ␣1,6-mannosylation: the former may be the same as the previously described core hexosylation of oyster hemocyte N-glycans (12), whereas the latter is more akin to fungal "outer chains," but the position of the ␣1,6-mannose is unusual in that it "replaces" the middle "B" ␣1,2-mannose that is removed first by the so-called ER mannosidase I. The most dominant N-glycan found in our analysis of the V. rubella N-glycome has a "humble" Man 2 GlcNAc 2 Fuc 1 structure, and, like a number of the smaller glycans in the margin snail, it lacks the "lower" arm ␣1,3mannose, in keeping with a greater degree of processing by mannosidases in "lower" eukaryotes than in vertebrates; the loss of the ␣1,3-mannose as observed here is known from insects and plants (38,52), while the bias in nematodes is toward removal of the ␣1,6-mannose (33).
Other than the standard core ␣1,3and ␣1,6-fucosylation, consistent with its status as an invertebrate and aforementioned core ␤1,3-mannosylation, complex extensions of the core ␣1,6-fucose are also observed in V. rubella: Not only was galactose linked to the fucose, but the galactose was often, in turn, substituted with ␣1,2-fucose and a zwitterion (see below). Such core modifications are not without precedent: Galactosylation of core fucose ("GalFuc") is known from cephalopod rhodopsins and nematodes (11,33), whereas Fuc␣1,2Gal␤1,4Fuc occurs in mutant Caenorhabditis strains (39,53). Other substitutions of the GalFuc motif include another galactose or a methylhexose as on keyhole limpet hemocyanin or in nematodes and Planaria (39, 54 -56); however, a zwitterionic modification of core GalFuc has never, to our knowledge, been previously described.
The zwitterionic glycans of the margin snail are generally modified with methylaminoethylphosphonate, but some examples of phosphorylcholine substitution were also found. Whereas both methylaminoethylphosphonate and aminoethylphosphonate are known as a modification of mollusc glycolipids (45,60) and aminoethylphosphonate is a component of jellyfish O-glycans (61) and protozoan glycolipid anchors (62), phosphorylcholine is a component of protein-and lipid-linked glycans from a range of species including nematodes, cestodes, annelids, and bacteria (63)(64)(65)(66). However, in V. rubella, the phosphorylcholine is linked to the galactose residue of the core GlcNAc modification and not to antennal GlcNAc as on the N-glycans of nematodes or cestodes. On the other hand, the closest "homologue" to the modification of N-glycans with methylaminoethylphosphonate is the presence of the nonmethylated form (aminoethylphosphonate) on both peripheral mannose and GlcNAc residues of the locust apolipophorin III (46). Thereby, the described linkages (both C6) on the locust glycans contrast to the probable mixture of substitutions to mannose, galactose, GlcNAc, and potentially GalNAc on both neutral and anionic N-glycan "backbones" of the margin snail.
Not only the range of modifications but the size of the N-glycans of V. rubella is noteworthy: We can propose structures with up to 20 residues (with masses of 3500 Da). This is probably not the upper limit (due to the limits of confident detection) but is of a range not typically associated with invertebrate glycans, albeit N-glycans of similar or larger masses are present in oysters (12), insects (37), and trematodes (67). This is due to the presence of both complex modifications and multiple branches, the latter probably being products of GlcNAc-TIV or V homologues catalyzing the formation of the triantennary structures found in these invertebrates.
O-glycans of V. rubella-The O-glycans detected by LC-ESI-MS in this study are, on the other hand, not so complex: We have detected either the O-GalNAc or O-Fuc types. The latter is represented by variants of GlcNAc␤1,3Fuc, which is the product of the fringe ␤1,3-N-acetylglucosaminyltransferase (68), but in contrast to insects (69), the "ionic" modification of O-fucose found in the margin snail is (methyl)aminoethylphosphonate rather than glucuronic acid. For the core 1 and core 3 type mucin-type O-glycans, fucose, sulfate, and (methyl)aminoethylphosphonate were among the modifications, the latter modification being akin to the linkage of aminoethylphosphonate directly to the reducing-terminal Gal-NAc found in jellyfish (61). There are few studies regarding the O-glycans of gastropods: Two examples, however, showed the presence of a bisubstituted core GalNAc on either a cone snail conopeptide toxin (70) or in whole snail tissues (71). This contrasts with only monosubstituted GalNAc being observed in the present study on V. rubella.
Analytical Considerations for Mollusc Glycomics-Key to all these analyses, especially of the N-glycans, was the differential release and solid-phase extraction yielding four pools prior to use of fused-core RP-HPLC. This reduced the number of glycans per fraction and so simplified the interpretation of fragmentation and digestion patterns as well as overcame suppression effects which may occur in complex mixtures of glycans of different types and amounts. As discussed in the introduction, many different procedures have been used to analyze mollusc glycomes in the past; here, we have brought together various methods used in different previously published and ongoing studies.
Porous and nonporous graphitized carbon was first used with mammalian glycans to desalt and separate neutral and sialylated glycans by elution with acetonitrile followed by acetonitrile supplemented with trifluoroacetic acid (72,73). C18based solid-phase extraction is usually used as a "clean-up" step, but not to separate types of glycans. Only recently was the ability of C18 to separate oligomannosidic from fucosylated N-glycans described, albeit under different conditions (74); independently, based on the concept that 10 -15% (v/v) methanol is required to elute N-glycans with non-and monosubstituted core ␣1,6-fucose from typical C18 RP-HPLC columns, we tested use of a step-gradient during solid-phase extraction. As we successfully used graphitized carbon to enrich sulfated and methylphosphorylated N-glycans from Dictyostelium away from the neutral structures (22), we adopted the approach of using NPGC and C18 in series prior to RP-HPLC. In the case of V. rubella, the glycans carrying sulfate or glucuronylated fucose were primarily in the "anionic" pool, whereas the "hydrophobic" pool is very much biased toward structures with core Fuc␣1,2Gal␤1,4Fuc␣1,6 motifs. The subsequent use of RP-HPLC facilitates the separation of isomeric and isobaric structures, a strategy established for some 30 years to resolve complex, paucimannosidic, and oligomannosidic glycans from mammalian and other sources (29).
Another aspect, which we consider important, is the optimized mix of methods on individual RP-HPLC fractions. The ability to "deconstruct" glycans using hydrofluoric acid and exoglycosidases, in addition to MS n , enables us to make further conclusions as to the composition and linkages, especially when dealing with unknown structures and lacking defined standards. Hydrofluoric acid has proven useful in the determination of linkages with phosphodiesters, phosphonates, galactofuranose, and fucose in previous studies from our own and other laboratories (22,27,45,(75)(76)(77), whereas linkage-specific glycosidases (such as the ␣1,6-mannosidase, ␣1,2-fucosidase, ␤1,4-galactosidase, and FDL ␤-hexosaminidase used here) are another tool for determining the nature of the isomeric status of the oligosaccharide struc-tures. Glycosidases and hydrofluoric acid indeed were key to determining the structure of, e.g. the methylaminoethylphosphonate-substituted N-glycans in this study as, based on MS spectra alone, these structures could have been easily misinterpreted as pauci-or oligomannosidic N-glycans. As one can expect that an organism should be able, as part of glycoprotein turnover in the lysosomes, to degrade its own glycans, it will be of interest to find further glycan-degrading enzymes capable of removing, e.g. glucuronic acid and methylhexose residues. CONCLUSION The present study exemplifies that there is, in comparison to insects or plants, no mollusc-specific glycomic characteristic signature. On the contrary, the majority of the N-and O-glycans described has for V. rubella have not been described for any other organism, although aspects of their structures (branched fucose, zwitterionic modifications, and elongated core modifications) have been found in other organisms (e.g. also in nematodes) or on other types of glycoconjugate (e.g. on glycolipids). This mix of highly modified glycans must have a biological function, which is difficult to elucidate considering the lack of genetic and microarray tools for gastropods in general. However, it is known that molluscs interact with other species in a glycan-dependent manner, e.g. Biomphalaria and Schistosoma or Crassostrea and Perkinsus (78,79), in a way that can involve a degree of glycomimickry. Considering that parasites often hijack systems of innate immunity, it may be that the unusual glycans of molluscs may have initially evolved as a defense mechanism in the absence of an adaptive immune response (which is present in vertebrates) or as a means for interactions with symbionts. The wide variability of mollusc glycomes may then reflect subtle differences in genetic and epigenetic systems enabling the evolution and adaptation of the "sugar" coating of their cells: As a specific glycan may be less essential than a specific protein, alteration of glycan modifications is a potential "red queen" strategy ("running to stay in the same place") for a group of species to survive (80). However, support for such a conclusion will need a more systematic set of analyses on related and less related mollusc species than has been performed to date.